[PATCH] x86/intel_rdt: Put group node in rdtgroup_kn_unlock

From: Jiri Olsa
Date: Tue Mar 14 2017 - 10:21:06 EST


The rdtgroup_kn_unlock waits for the last user to release
and put its node. But it's calling kernfs_put on the node
which calls the rdtgroup_kn_unlock, which might not be
the group's directory node, but another group's file node.

This race could be easily reproduced by running 2 instances
of following script:

mount -t resctrl resctrl /sys/fs/resctrl/
pushd /sys/fs/resctrl/
mkdir krava
echo "krava" > krava/schemata
rmdir krava
popd
umount /sys/fs/resctrl

It triggers the slub debug error message with following command
line config: slub_debug=,kernfs_node_cache.

Calling kernfs_put on the group's node.

Signed-off-by: Jiri Olsa <jolsa@xxxxxxxxxx>
---
arch/x86/kernel/cpu/intel_rdt_rdtgroup.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/x86/kernel/cpu/intel_rdt_rdtgroup.c b/arch/x86/kernel/cpu/intel_rdt_rdtgroup.c
index c05509d38b1f..9ac2a5cdd9c2 100644
--- a/arch/x86/kernel/cpu/intel_rdt_rdtgroup.c
+++ b/arch/x86/kernel/cpu/intel_rdt_rdtgroup.c
@@ -727,7 +727,7 @@ void rdtgroup_kn_unlock(struct kernfs_node *kn)
if (atomic_dec_and_test(&rdtgrp->waitcount) &&
(rdtgrp->flags & RDT_DELETED)) {
kernfs_unbreak_active_protection(kn);
- kernfs_put(kn);
+ kernfs_put(rdtgrp->kn);
kfree(rdtgrp);
} else {
kernfs_unbreak_active_protection(kn);
--
2.9.3